Oscar L.

Need a Python developer who builds systems your team can actually trust with real data? That's what I do. I'm a Python Data Engineer with 2+ years building production systems end to end — from ingestion and processing to the final dashboard a director opens every morning. Most of my work now involves LLMs, and my approach is simple: code computes what can be computed, and the model only handles judgment. On a recent project for a German software client — PDF orders for custom-cut glass, where one misread measurement scraps the product — the system verifies every extracted value against the source and returns an empty field before an invented one. That discipline is measurable, not a slogan. On one document I ran 41 readings across 6 layouts to find the real failure rate, then built the check that catches it. Zero false positives. What I do: - Data pipelines end to end — Apache Airflow, NiFi, Kafka, RabbitMQ, MQTT, from ingestion through to what the business reads - Python backends and APIs — FastAPI, REST, integrations, authentication, webhooks - AI agent workflows — directing Claude Code and Cursor, and critically reviewing what they produce before it ships - Document and PDF extraction with schema validation, verbatim verification and confidence gating - Apache Superset — my specialty: setup, security, roles, datasets, performance tuning, deployment - PostgreSQL design and query optimization Some background: I spent 18 months on a production emergency-response platform — sensor and drone metadata normalized into strict JSON contracts, hundreds of thousands of records, 30+ Superset dashboards on top, demoed live to government auditors. On a recent healthcare project the wins that mattered were on performance — rewriting queries so dashboards loaded fast enough to be used live in clinical settings. I keep communication clear and fast (English C1, native Spanish), and I care more about the system being used than about it being impressive. Tell me what you're working with and what it needs to do — I'll tell you honestly how I'd approach it.